mycoscience 54 (2013) 75e86

Available online at www.sciencedirect.com

ISSN 1340-3540 (print), 1618-2545 (online)

journal homepage: www.elsevier.com/locate/myc

Review Origin and evolution of the powdery mildews (, )

Susumu Takamatsu*

Department of Bioresources, Graduate School, Mie University, 1577 Kurima-Machiya, Tsu 514-8507, Japan article info abstract

Article history: Molecular phylogeny suggests a close relationship of Asteraceae to the early evolution of Received 2 August 2012 Golovinomyces. The family Asteraceae, with a geographic origin in South America, expanded Accepted 15 August 2012 into the Northern Hemisphere, where it may have been infected by an ancestor of Golo- Available online 29 September 2012 vinomyces, thus starting a close hosteparasite relationship. Using this event as a calibration point, we designed molecular clocks for powdery mildews using the 28S rDNA D1/D2 and Keywords: internal transcribed spacer (ITS) regions. According to these clocks, the powdery mildews Biogeography originated in the Late Cretaceous and the first radiation of the major lineages occurred at Erysiphaceae the Cretaceous/Paleogene boundary. Ancestral powdery mildews may have first radiated Molecular clock on broad-leaved deciduous trees in the high latitudes of the Northern Hemisphere, and Molecular phylogeny continued further speciation whilst migrating to southward during the world cooling in the Paleogene Paleogene and Neogene periods. The cradle of four herb infecting genera, viz. Blumeria, Golovinomyces, Leveillula, and Neoe¨rysiphe may be within the area extending from Central/ West Asia to the Mediterranean. ª 2012 The Mycological Society of Japan. Published by Elsevier B.V. All rights reserved.

1. Introduction or Ferns (Amano 1986). Of the total host species, 9176 are dicotyledons, the remaining 662 being monocotyledons. Of Powdery mildews are an important group of plant pathogenic the monocotyledons, 634 species belong to the subfamily fungi forming a white, powdery film on leaves, stems, flowers, Pooideae of Poaceae (the grass family). and fruits of Angiosperms. Sixteen genera and ca 900 species Several scientific speculations on phylogenetic relation- are presently known worldwide (Braun and Cook 2012). All of ships and evolution of powdery mildews based on the species are exclusively obligate biotrophs of plants. morphology and host relationships have been published Molecular phylogenetic analyses so far reported indicate that (Neger 1901; Arnaud 1921; Raymond 1927; Blumer 1933; the powdery mildews form a distinct monophyletic group Katumoto 1973; Braun 1987, 1995). Most of these arguments (Mori et al., 2000a,b; Wang et al., 2006a,b). They may be simply devolved around characteristics thought to be primi- a fungal group that acquired an obligate biotrophic nature tive or derived. Some critical characteristics are as follows: only once in their ancestry and have retained it ever since. Up (1) number of asci in an ascoma, e.g. one or several; (2) number to 9838 plant species have been recorded as hosts of powdery of ascospores in an ascus, e.g. eight or fewer; (3) morphology mildews, all of which are Angiosperms but not Gymnosperms of appendages; (4) conidiogenesis, conidia maturing in chains,

* Tel.: þ81 59 231 9497; fax: þ81 59 231 9540. E-mail address: [email protected]. 1340-3540/$ e see front matter ª 2012 The Mycological Society of Japan. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.myc.2012.08.004 76 mycoscience 54 (2013) 75e86 catenescent (Euoidium-type) or maturing one at a time (Pseu- 2.2. Speculation on evolutionary dating in the absence of doidium-type); (5) mycelium, ectotrophic or endotrophic. Of a molecular clock these, most scientists consistently regarded eight-spored asci and mycelioid appendages as primitive characters. These Since all members of powdery mildews are obligate biotrophs arguments were reviewed in Braun (1987, 1995) and Braun and of plants and their hosts are restricted to Angiosperms, the Cook (2012). appearance of a typical powdery mildew cannot pre-date the Hirata (1966) and Amano (1986) constructed a comprehen- Angiosperms. Also, since the monocotyledonous hosts are sive, worldwide list of powdery mildews and their host plants mostly restricted to the subfamily Pooideae of Poaceae, it is based on nearly 4000 references. It consists of 14,647 powdery likely that a dicot infecting powdery mildew spread to an mildew/host combinations, and the countries where they ancestor of Pooideae after the monocot/dicot split. From these were recorded (Amano 1986). This list has been utilized by assumptions it may be possible to speculate the earliest origin many scientists as a comprehensive database of powdery of powdery mildews. mildews and their hosts. The biogeography, phylogeny and The origin of Angiosperms based on the molecular clock is evolution of the powdery mildews were discussed with not congruent with the estimates of fossil ages. Calibration by reference to the data (Hirata 1968, 1969, 1971a,b, 1972, 1975, molecular clock suggests Angiosperms first appeared in the 1976, 1980; Amano 1986, 1992, 2002). These papers provide Early-Middle Jurassic Period [179e158 Ma] (Wikstro¨m et al. a unique insight into these subjects because other scientists 2001). On the other hand, fossil records indicate that Angio- have mostly based such arguments solely on morphology. sperms first appeared in northern Gondwana during the Early Since previous studies on molecular phylogeny of the Cretaceous Period, approximately 135 Ma (Barrett and Willis powdery mildews have already been reviewed in Braun et al. 2001). The oldest fossil of a monocot was found in the (2002), Takamatsu (2004), Glawe (2008) and Braun and Cook United States from Late Cretaceous sediments [90 Ma] (2012), this review is focused on two other aspects, viz. (Gandolfo et al. 1998). The molecular clock suggests that evolutionary dating and geographic origin of the powdery monocots have originated in Late Jurassic (Wikstro¨ m et al. mildews. 2001). Based on these reports, it is unlikely that powdery mildews originated earlier than the Jurassic Period.

2. Evolutionary dating of powdery mildews 2.3. Design of molecular clocks for powdery mildews

The evolution of an organism is mostly driven by the Earth’s Mori et al. (2000b) determined the nucleotide sequences of the environmental changes and the concurrent evolution of other 18S and 28S rDNA for ten powdery mildew species and used organisms. Host plants are an absolute prerequisite for the them to determine the phylogenetic placement of powdery survival of an obligate biotroph like a powdery mildew. So we mildews in the Ascomycota. They used the molecular clocks cannot discuss the evolution of powdery mildews without of 18S rDNA reported by Simon et al. (1993) (0.667% per lineage relating them to the evolution of their hosts. Thus Earth’s per 100 Ma) and Berbee and Taylor (1993) (1.0% per lineage per environmental changes influence the evolution of powdery 100 Ma). Simon et al. (1993) used the monocot/dicot split mildews either directly or indirectly via their effects on host [200 Ma] as a calibration point. However, subsequent studies plants. Therefore, an estimation of the date of evolutionary estimate the split to have occurred later than 200 Ma events is essential to discuss the evolution of powdery (Gandolfo et al. 1998; Wikstro¨ m et al. 2001). Thus, the molec- mildews in relation to their hosts and environmental factors. ular clock of Simon et al. (1993) may be too slow. It is note- worthy that Berbee and Taylor used several fossil records as 2.1. Fossil record of powdery mildews calibration points for their molecular clock that was later changed to 1.26% per 100 Ma based on a re-assessment of Whilst fossil records are commonly used as direct evidence of fossil records (Berbee and Taylor’s 2001), and so their molec- evolution, they are relatively rare in fungi compared with ular clock may be more reliable than that of Simon et al’s. 1993 those of animals and plants and there is no reliable fossil clock. Berbee and Taylor’s 1993 molecular clock and 18S rDNA record, especially for powdery mildews (Braun and Cook sequences estimated the split of powdery mildews from 2012). According to a summary of fungal fossils by Tiffney Myxotrichaceae (putative sister group of powdery mildews) as and Barghoorn (1974), so far four records have been listed, of 127 24 Ma, and the first divergence within the powdery which the most reliable, named Protoascon missouriensis was mildews as 92 24 Ma, both of which suggest powdery discovered in the mid Pennsylvanian Epoch of the Permian mildews originated in the Cretaceous Period. Berbee and Period in strata laid down ca 300 million years ago (Ma) (Batra Taylor’s 2001 clock (1.26% per 100 Ma) resulted in the split et al. 1964). However, subsequent studies re-interpreted this and divergence dates of 101 and 73 Ma, respectively, still record as a kind of zygomycete (Pirozynski 1976; Taylor suggesting a cretaceous origin of powdery mildews (Fig. 1). et al. 2005). Stubblefield and Taylor (1983) re-investigated the The maximum genetic divergence of the 18S rDNA within fossil Traquairia found in coal balls of the same epoch, and the powdery mildews was only 1.85% and the maximum reported that it is most similar to the Eurotiales or Erysiphales. substitutions were only 37 bases. This rate of change of 18S However, as far as I can see from the photo, there is no strong rDNA may be too slow to calibrate evolutionary dating within evidence to consider it a member of the Erysiphales. A powdery mildews. Furthermore, Berbee and Taylor’s 2001 comprehensive recent overview on fossil fungi was published clock did not include fossil records of powdery mildews, and by Taylor and Krings (2005). there is no evidence that their 18S rDNA evolves at the same mycoscience 54 (2013) 75e86 77

asteraceous hosts with Golovinomyces in early evolution. The branching order of the Golovinomyces tree was similar, although not identical, to the branching order of the tribes of the Asteraceae, suggesting co-speciation between Golovino- myces and Asteraceae. The geographic origin of the Asteraceae was in South America from whence they expanded into the Northern Hemisphere. All the asteraceous hosts used in Matsuda and Takamatsu (2003) belong to tribes that diverged in the Northern Hemisphere. It is possible that the ancestor of Golovinomyces first infected asteraceous hosts in South Amer- ica making this region the origin of the genus. There is a report that Oidium mutisiae, being a Golovinomyces-like anamorph, occurs on two Mutisia species of the tribe Mutisieae of the Fig. 1 e Estimated dates of origin and divergence of the Asteraceae in Argentina (Havrylenko 1993). Mutisieae is a tribe powdery mildews based on the nucleotide sequences of mainly distributed in South America, but is sister to a group of the 28S rDNA D1/D2 region and 18S rDNA. A is the splitting tribes that radiated in the Northern Hemisphere (Bremer 1994). of the Erysiphales and Myxotrichaceae, and B is the first If an ancestor of Golovinomyces infected an asteraceous host radiation within the Erysiphales. Ma: million years ago. before the Asteraceae expanded into the Northern Hemi- sphere, the Oidium on Mutisia should occupy the very base of the Golovinomyces tree. However, the phylogenetic analyses rate as other fungal groups. Therefore, it was necessary to gave us an unexpected result (Takamatsu et al. 2006). The design a molecular clock of DNA regions that evolve faster powdery mildew isolates on Mutisia were not placed at the base than 18S rDNA, i.e. the variable region of 28S rDNA or internal of Golovinomyces. Instead, they nested in clades consisting of transcribed spacer (ITS) region, using a revised calibration isolates from the Northern Hemisphere and furthermore, they point for powdery mildews. separated into two different clades, indicating that the isolates A molecular clock needs at least one calibration point and from each Mutisia species were derived from different ances- there are three options for obtaining it using: 1) fossil records, tors. This result suggests that an ancestor of Golovinomyces 2) geological events like the breakup of a continent, orogeny, occurring in the Northern Hemisphere first infected Aster- the appearance of a big river, etc., 3) co-speciation of hosts and aceae after the family had migrated to the Northern Hemi- parasites. The first option may be impossible at present sphere and before it split off the Cardueae, the first tribe because there is no reliable fossil record of powdery mildews as radiated in the Northern Hemisphere (Takamatsu et al. 2006). mentioned above. Whilst the second option may be effective Then Golovinomyces may have migrated from the Northern into for organisms lacking long distance mobility like Damaster the Southern Hemisphere and infected Mutisia on two separate ground beetles (Su et al. 1998), powdery mildews may travel occasions. If the timing of the divergence of the Cardueae was longer distances by means of air-borne spores. Although it is known, it would provide a calibration point for the design of not clear exactly how efficient long distance dispersal is with a molecular clock for powdery mildews. powdery mildews, there is a report that the cereal powdery Unfortunately, there was no report of the timing of this mildew, Blumeria graminis, has flown about 650 km from divergence. Therefore, we used the molecular clock of the rbcL England to Denmark (Hermansen et al. 1978). In the United gene (Bremer and Gustafsson 1997), and this suggested that States, conidia of Golovinomyces cichoracearum on lettuce flew the Cardueae split from other asteraceous tribes 25.2 Ma 190 km northward from Salinas Valley, California (Schnathorst (Takamatsu and Matsuda 2004). This result was congruent with 1959). Therefore, there is some doubt about the validity of the report of Kim et al. (1998) that the Asteraceae underwent an using the second option due to powdery mildews being able to explosive radiation and dispersal during or near the transition surmount certain physical barriers. The third option as used between the Oligocene and Miocene Epochs [ca 23 Ma]. We for the relationships with long-term co-speciation such as used this date in the Golovinomyces trees constructed by insects and their endosymbiotic bacteria (Moran et al. 1993) sequences from the ITS region including the 5.8S rDNA, and the may, however, be applicable. In powdery mildews the first 50 end of the 28S rDNA including the D1 and D2 regions candidate for studies on co-speciation was B. graminis and its (Takamatsu and Matsuda 2004). When the node of the first split grass hosts (subfamily Pooideae). As stated above, the present was dated 25.2 Ma, the nucleotide substitution rates were form of B. graminis probably originated after the first appear- found to be 2.52 0.11 10 9 per site per year [0.01D ¼ 3.97 Ma; ance of the grasses, i.e. due to infection of an ancestor of the D ¼ genetic distance calculated by the Kimura’s two-parameter subfamily. If co-speciation between B. graminis and its hosts model (Kimura 1980)] in the ITS region and 6.5 0.4 10 10 per started after this infection, it may be possible to estimate site per year (0.01D ¼ 15.4 Ma) in the D1 and D2 regions of the calibration points based on this relationship. However, this 28S rDNA (Takamatsu and Matsuda 2004). Using this clock, date approach proved difficult as Japan has only a few grass hosts. of the first radiation of the major tribe was estimated to be ca 70 Meanwhile, phylogenetic analyses of Golovinomyces revealed Ma (Fig. 1). Based on complete sequences of the 18S rDNA and that all clades situated at the base of the tree were occupied by the molecular clock (1.26% per 100 Ma) reported by Berbee and isolates from asteraceous hosts, and isolates from single tribes Taylor (2001), the first radiation of the major tribe was dated ca of Asteraceae occupied several basal clades (Matsuda and 73 Ma, which agrees well with the date calculated by our Takamatsu 2003). This result suggests a close affinity of molecular clock. Oberhaensli et al. (2011) calculated the 78 mycoscience 54 (2013) 75e86

emergence of B. graminis on wheat and barley using a molecular chasmothecium was introduced for ascomata of powdery clock of a transposable element. The result was in good mildews (Braun et al. 2002; Braun and Cook 2012). Sugiyama agreement with our calculation. et al. (1999) reported that the Myxotrichaceae has its own lineage distantly related to the other onygenalean fungi and 2.4. Evolutionary dating using molecular clocks of closely related to the Leotiales and Erysiphales. Mori et al. powdery mildews (2000b) confirmed the report of Sugiyama et al. (1999) by using newly determined 18S and 28S rDNA sequences from Calculations of evolutionary dating using the molecular clocks 10 powdery mildew species. Wang et al. (2006a,b) reported of 28S rDNA (Takamatsu and Matsuda 2004) and 18S rDNA that Chlorociboria (Helotiaceae) and Cyttaria (Cyttariaceae) are (Berbee and Taylor 2001) suggested that the split of powdery the closest relatives of powdery mildews based on combined mildews and Myxotrichaceae occurred in the Middle Creta- data of 18S, 28S and 5.8S rDNA. Peterson and Pfister (2010) ceous Period [ca 100 Ma] and first split within powdery conducted a phylogenetic analysis of the mildews occurred in the Late Cretaceous [90e80 Ma] including powdery mildews using combined data of five DNA (Takamatsu 2004; Takamatsu et al. 2005). This dating is clearly regions including EF1-alpha gene. In this analysis, Cyttaria and later than the oldest angiosperm fossils [ca 135 Ma], support- Chlorociboria were near to the powdery mildews but not the ing the speculation that powdery mildews infected Angio- closest relatives. Instead, Myxotrichaceae and Pleuroascus sperms after they split from the Gymnosperms. Radiation of (Pseudeurotiaceae) proved to be the closest relatives. the five major lineages of the Erysiphaceae occurred in the Cyttaria, an obligate pathogen of plants, has Nothofagus, Cretaceous/Paleogene boundary [70e58 Ma] (Takamatsu and a Gondwanan plant, as host and is distributed only in the Matsuda 2004). Radiation of herb-parasitic powdery mildews Southern Hemisphere. The species of Cyttaria have been in the began in the Miocene Epoch of the Neogene Period with the focus of evolutionary biologists since Charles Darwin in terms first radiation within Golovinomyces and Neoe¨rysiphe [25e10 Ma] of co-evolution with Nothofagus and their biogeography (Takamatsu et al. 2008a). This was followed by the formation (Peterson et al. 2010). Although it is interesting that powdery of the subsection Magnicellulatae (formerly Sphaerotheca fuligi- mildews are closely related to Cyttaria, it may be difficult to nea)ofPodosphaera section Sphaerotheca in the Middle Miocene find direct morphological affinities between Cyttaria and [15e10 Ma] (Takamatsu et al. 2010), and by the first split within powdery mildews. Since Chlorociboria is a fungal group to Leveillula in the Pliocene [5 Ma] (Takamatsu et al. 2008b). produce typical apothecia, it is also difficult to find common The global cooling of the Earth and an increase in the arid morphological characteristics with powdery mildews. On the area followed Alpine and Himalayan orogenies caused an other hand, both Myxotricaceae and Pleuroascus produce increasing area of steps containing grasses and asteraceous cleistothecia (ascomata without ostiole) similar to the chas- herbs (Tiffney 1985a,b; Tiffney and Manchester 2001; mothecia of powdery mildews. In addition, some members Takahashi 2006). This may have triggered the transfer of of Myxotrichum have appendages with circinate tips powdery mildews to herbaceous plants and the divergence of (Alexopoulos et al. 1996), which are similar to the uncinuloid these herb-parasitic species. Thus, the estimation of evolu- appendages of some powdery mildews. Members of the tionary dating using the molecular clock suggests that the Myxotricaceae have cellulolytic characters and colonize plant evolution of powdery mildews is closely related to the changes debris (Sugiyama et al. 1999; Tsuneda and Currah 2004). It is in the Earth’s environment and its flora. known that powdery mildews also encode cellulase genes (Pryce-Jones et al. 1999). Thus, it might be possible that an ancestral fungus colonizing on plant debris has obtained 3. Origin of powdery mildews biotrophic ability, leading to obligate parasitism. Tsuneda and Currah (2004) reported that ascomata of Myxotrichum arcticum 3.1. Closest relatives of powdery mildews bear a striking resemblance to apothecia in morphogenesis, supporting the derivation of the Myxotrichaceae from a hel- A survey of closest relatives is important for speculation on otialean ancestor. Detailed morphogenetic observations of the origin of powdery mildews. Powdery mildews have powdery mildews might support the results of molecular previous often been placed in Plectomycetes based on their phylogenetic analyses. cleistothecial ascomata, i.e. closed fruiting bodies without ostiole (Ainsworth et al. 1971; Webster 1980). On the other 3.2. Geographic origin of powdery mildews hand, the production of asci from a basal hymenium and the forcible discharge of ascospores have been used to classify Publications discussing the geographic origin of powdery powdery mildews as Pyrenomycetes (Yarwood 1973, 1978; mildews are very rare. Heluta (1992) speculated southern Alexopoulos and Mims 1979). Saenz et al. (1994) pointed out China as the geographic origin. Unfortunately, I do not have that powdery mildews belong to neither Pyrenomycetes nor sufficient data to evaluate his speculation. Thus, it is at Plectomycetes based on phylogenetic analysis of 18S rDNA present only possible to speculate on geographic origin of sequences. Instead, they reported a close relationship powdery mildews below. between B. graminis and Sclerotinia sclerotiorum (Leotiales). Based on closed, non-ostiolate ascomata as in cleistothecia, 3.2.1. Physical geography, climate and flora in Tertiary but with asci arranged in hymenial arrangement as in (Paleogene and Neogene periods) perithecia as well as the phylogenetic position distant Physical geography, climate and flora in the Tertiary Period of true Plectomycetes and Pyrenomycetes, the new term (65.5e2.6 Ma, now split into Paleogene and Neogene periods of mycoscience 54 (2013) 75e86 79

the Cenozoic Era), are described below using the following Manchester 2001). The latitude of the BLB in the Paleogene publications: Tiffney (1985a,b; 2000); Wen (1999); Manchester Period [65.5e23 Ma] was approximately 75N, then moving to (1999); Tiffney and Manchester (2001); Takahashi (2005, its present latitude (ca 65N) as the Paleogene progressed and 2006). These references can be consulted to obtain more North America rotated counterclockwise. During the Miocene detailed information. Epoch [23e5.3 Ma] of the Neogene Period, the BLB was suitable The Mesozoic/Cenozoic boundary [65.5 Ma] is well known for the exchange of temperate deciduous plants, such as as the point of extinction of the dinosaurs. However, many Castanea, Juglans, Magnolia, and Quercus, and remained avail- other organisms also became extinct at this time. The Ceno- able for floristic exchanges until about 3.5 Ma (Wen 1999; zoic Era is the period of mammals and Angiosperms. It begins Tiffney 2000; Tiffney and Manchester 2001). with the Paleogene Period containing three successive epochs, The North Atlantic land bridge (NALB) was connected to Paleocene, Eocene and Oligocene. The Paleocene Epoch star- Europe in the Paleogene Period via Greenland, with a northern ted with a warm climate, which continued to the Early Eocene connection in the high Arctic region and a southern connec- Epoch [ca 50 Ma]. Tropical and subtropical flora extended to tion through southern Greenland. The latitude of the southern lat. 65N. Broadleaf deciduous trees like the genera Alnus, route was approximately 60N, thus further south than the Betula, Quercus, Juglans, Populus, and Acer were distributed in BLB. Floristic migration via the NALB was possible during the the Arctic area at latitude above 70N(Takahashi 2006). Paleocene and Eocene Epochs [65.5e34 Ma], the next epoch in A climate deterioration involving a decrease in world the Paleogene Period (Wen 1999; Tiffney 2000; Tiffney and temperature and an increase in seasonality occurred during Manchester 2001). the Late Eocene [37.2e33.9 Ma] or at the EoceneeOligocene The Turgai Straits separated Europe from Western Siberia boundary [33.9 Ma] (Tiffney 1985b). Thermophilic and ever- from the mid-Mesozoic Era to the end of the Eocene Epoch [ca green taxa became less common and restricted to favourable 34 Ma] in the Paleogene Period and are generally presumed sites in much of Europe and North America. Temperate forests to have formed a biogeographic barrier to animals (Tiffney with evergreen and deciduous trees were distributed in Eur- 1985b; Tiffney and Manchester 2001). It clearly separated asia, North America and North Africa in the Oligocene European and Asian Faunas until its demise in the Oligocene [33.9e23.0 Ma] (Takahashi 2005). The genera Carya, Liquid- [34e23 Ma]. ambar, Cercidiphyllum, Alnus, Corylus, Carpinus, Nyssa, Quercus, and Ulmus were the major components of the temperate 3.2.2. Northern Hemisphere origin or Southern Hemisphere forests. origin? Towards the boundary with the Neogene Period which When discussing the origin of powdery mildews, it is important contains the Miocene and then the Pliocene Epochs, a gradual to consider the geographic origin of Angiosperms due to the warming that extended into the Early Miocene, led to the close host/parasite relationships. Although there are several expansion of several evergreen and thermophilic lineages in different hypotheses as to the origin of Angiosperms, North Europe and coastal margins of North America. Starting in the Gondwana in the Early Cretaceous Period (ca 135 Ma) may be middle Miocene [ca 15 Ma], cooling returned and progressed the most likely (Barrett and Willis 2001; Takahashi 2006). stepwise into the Quaternary Period containing the Pleisto- The genus Uncinula (now sect. Uncinula) has cene Epoch and the ice ages (Tiffney and Manchester 2001). a relatively old origin among powdery mildew genera (Mori This led to sequential modernization of the flora in Europe and et al. 2000a). In the monograph of Braun (1987),86Uncinula North America, involving the spread of deciduous trees and species are listed. Most of them (54 species) are distributed in herbs and the loss of thermophilic, evergreen elements East Asia and only 10 in South America, of which 8 are (Tiffney and Manchester 2001). Back in the Miocene Epoch of endemic to South America. Four species of Erysiphe sect. the Neogene Period in Western and Central Asia, a temperate Uncinula have been recorded on Nothofagus (Braun 1987; flora evolved at the southern end of the Turgai Depression Havrylenko and Takamatsu 2003; Meeboon and Takamatsu near the Paratethys, while to the east in Kazakhstan, a series 2012), which is a typical Gondwanan genus, having a rela- of great lakes developed, surrounded by mesic (moderately tively old origin within broad-leaved deciduous trees. Three of moist) vegetation of Pterocarya, Betula, Alnus, Salix, Nyssa, and the Erysiphe sect. Uncinula species parasitic to Nothofagus, so forth. Adjacent Alpine and Himalayan orogenies shut off E. nothofagi, E. patagoniaca (Havrylenko and Takamatsu 2003), the southern source of moisture, allowing these areas to and E. havrylenkoana (Meeboon and Takamatsu 2012) have become arid, further isolating forest communities and unique appendages spiral coiling like a spring, which are contributing to the change in climate and vegetation in unknown in powdery mildews of the Northern Hemisphere. Central Asia and to the north (Tiffney and Manchester 2001). Assuming genuine co-speciation between Angiosperms and In the central and northern Turgai Depression, the diversity of powdery mildews, these three species are candidates for an Angiosperms decreased during the Miocene Epoch. Open old origin. To address this possibility, Niinomi et al. (2008) vegetation began to appear in the area by the Early Miocene determined rDNA sequences for the Nothofagus powdery [23e16 Ma], and by the Early Pliocene Epoch [5.3e3.6 Ma], the mildews. The sequences were compared in a phylogenetic area was a mosaic of deciduous forests (Tilia, Quercus, Carpi- analysis with those from Northern Hemisphere species of nus, Cotylus) and open steppe dominated by grasses, Aster- Erysiphe sect. Uncinula. Surprisingly, the sequences from the aceae, and Chenopodiaceae (Tiffney and Manchester 2001). powdery mildews on Nothofagus were not placed at the base of The Bering land bridge (BLB) has connected north-eastern the tree, instead they were closely related to those of the Asia and north-western North America several times since Northern Hemisphere (Niinomi et al. 2008), indicating that the Mesozoic Period (Wen 1999; Tiffney 2000; Tiffney and they are not ancestral in the Erysiphales. An ancestor of the 80 mycoscience 54 (2013) 75e86

species may have infected Nothofagus after migrating into appendages. Thus, among the tree-parasitic taxa, those South America from the Northern Hemisphere after the having uncinuloid appendages may be the most ancestral. breakup of the Gondwana continent. This possibility is sup- In the tribe Cystotheceae, Sawadaea, which also has unci- ported by the fact that Uncinula-like Erysiphe species on nuloid appendages, occupies the basal position. So, Acer- Nothofagus are reported only from South America, and not aceae, a main host family of Sawadaea, may be one of the early from Australia, New Zealand, or New Guinea. The co- host families of powdery mildews. The base of the tribe evolution of Golovinomyces and Asteraceae described above Phyllactinieae is occupied by Pleochaeta and Queirozia, both also does not support a Southern Hemisphere origin of Golo- with uncinuloid appendages. Therefore, Ulmaceae, a main vinomyces as well (Takamatsu et al. 2006). host family of Pleochaeta, may be an additional early host Hirata (1972) surveyed host species of powdery mildews family. Although the host family of Queirozia is Fabaceae, it is and their distribution for 167 angiosperm families excluding not likely that this represents an early host family of powdery small families composed of less than 30 species and the mildews. Aceraceae and Ulmaceae are also the main host families which have special nutrient requirements, and/or an families of Erysiphe sect. Uncinula (previously the genus Unci- ecological habit making them parasitic, insectivorous or nula). This section comprises hosts belonging to 39 host plant aquatic. It was shown that all 13 families distributed only in families (Amano 2002), most of them are members of the the Northern Hemisphere were host to a powdery mildew. families Salicaceae, Betulaceae, Fagaceae, Moraceae, Acer- Moreover, the most popular host families, the Salicaceae, aceae, Vitaceae, and Anacardiaceae. In addition, taxa occur- Betulaceae, Brassicaceae, Geraniaceae, and Aceraceae, ring on Liquidambar (Hamamelidaceae) and Lagerstroemia belonged to this distribution type. Of the 52 families endemic (Lythraceae) occupy the very base of Erysiphe sect. Uncinula.It to tropical areas, only 23 (44%), were hosts and of the seven is noteworthy that all involved plant families and/or genera families endemic to the Southern Hemisphere, only two (29%) encompass broad-leaved deciduous trees. were hosts. When a similar investigation was done with Calibration by molecular clock suggested that powdery species of plants belonging to families confined to the mildews originated in the Late Cretaceous Period and the major Northern Hemisphere, 11.8% were hosts. By contrast, only lineages diverged at the Cretaceous/Paleogene Period boundary 0.4% and 0.1% of the species confined respectively to tropical [70e57 Ma] (Takamatsu and Matsuda 2004). The temperature of areas and the Southern Hemisphere were hosts. Therefore, the Earth was warmer than present time from this boundary to powdery mildews frequently infect plants found in the Early Eocene [ca 50 Ma], and broad-leaved deciduous trees, the Northern Hemisphere, but are less frequent on tropical plants early hosts of powdery mildews, were distributed in the Arctic and those in the Southern Hemisphere. Taking into consid- area at a latitude higher than 70N(Tiffney 1985b; Manchester eration these uneven distributions of powdery mildews and 1999; Tiffney and Manchester 2001; Takahashi 2005, 2006). the papers cited above, a Southern Hemisphere or tropical Therefore, it is likely that the early divergence of powdery origin of powdery mildews seems unlikely. mildews occurred in this region. A climatic deterioration involving a decrease in world temperature and an increase in 3.2.3. Cradle of powdery mildews seasonality started to occur from the Middle Eocene [ca 50 Ma] Because powdery mildews are obligate biotrophs of plants, and continued until the Glacial Ages in the Pleistocene Epoch of their niche is strictly confined to the surface of living plants. the Quaternary Period [ca 2.6 Ma]. This may have resulted in Geographic distributions of the early hosts of powdery a southward migration of deciduous trees together with their mildews in the Paleogene Period when their first radiation associated powdery mildews. These climatic as well as occurred may provide a base to speculate on the location of geological changes in the Neogene Period may have triggered the cradle of these pathogens. Powdery mildews are divided the further divergence of powdery mildews. into five large lineages (tribes) and two basal genera, Para- uncinula and Caespitotheca (Mori et al. 2000a). Of the five tribes, 3.2.4. The basal genera Parauncinula and Caespitotheca three, viz. Erysipheae, Phyllactinieae, and Cystotheceae, Parauncinula is a small genus consisting of only two species, include both tree-parasitic and herb-parasitic taxa, whereas P. septata and P. curvispora. Both species infect Fagaceae and the remaining two tribes, viz. Blumerieae and Golovinomy- their distributions are restricted to East Asia (Japan and ceteae, include only herb-parasitic taxa with only a few China). The monotypic genus Caespitotheca is only known in exceptions. In the first-mentioned three tribes, tree-parasitic tropical Argentina on Schinopsis (Anacardiaceae) (Braun and taxa usually occupy basal positions in the lineages and herb- Cook 2012). One possible scenario to explain why these two parasitic species a higher (derived) position. In addition, basic genera are isolated in narrow areas of two different both the two totally basal genera are tree-parasitic. This continents may be as follows. Ancestors of these genera evidence suggests that powdery mildews were originally tree- originated in high latitudes of the Northern Hemisphere. They parasitic and subsequently expanded their host ranges to then accompanied the southward migrations of their hosts herbaceous plants (Takamatsu et al. 2000; Mori et al. 2000a; and having survived many extinction events, Parauncinula and Takamatsu 2004). This conforms well to the evolutionary Caespitotheca now remain as relict distribution in East Asia and history of Angiosperms (Tiffney and Manchester 2001; Argentina, respectively. Takahashi 2006). Among tree-parasitic taxa, those with teleomorphs (chasmothecia) having uncinate-circinate 3.2.5. Comparisons of powdery mildew distributions among appendage tips (uncinuloid type) usually occupy the very Europe, North America, Central & West Asia, and East Asia base of the respective lineages, and the two basal genera, Hirata (1971b) divided the Northern Hemisphere into four Parauncinula and Caespitotheca, also have uncinuloid regions, viz. Europe, North America, East Asia, and Central & mycoscience 54 (2013) 75e86 81

West Asia, and counted the number of powdery mildews which occur in two of the regions but are not found in the other two regions. He found that the distribution in Europe was closely related to that in North America and Central & West Asia, but not to that in East Asia. However, the distri- bution in North America was somewhat related to East Asia (Fig. 2). From these results, he concluded that the distribution in East Asia is unique in the Northern Hemisphere and that of North America is intermediate between Europe and East Asia. This uneven distribution of powdery mildews in the Northern Hemisphere may be partly explained by the formation of Turgai Strait, the NALB and BLB. Until end of the Paleogene Period, Europe and East Asia were separated by the Turgai Fig. 3 e Phylogenetic tree of Neoe¨rysiphe except for N. Strait which may have barred the exchange of powdery galeopsidis. European group split from the North & South mildews and their host plants. On the other hand, the NALB American group. Then, East Asian group split from a part and BLB may have allowed the exchange of powdery mildews of the North & South American group. between Europe and North America, and between North America and East Asia, respectively. Heluta et al. (2010) conducted molecular phylogenetic analyses of Neoe¨rysiphe. Firstly, a N. galeopsidis clade split off (Fig. 2). On the other hand, herb-parasitic taxa are more from a clade composed of other Neoe¨rysiphe species. The latter common in Europe and Central & West Asia than in East Asia. clade then split into a clade of Neoe¨rysiphe spp. confined to These two main groups based on host ranges, viz. the tree- Europe and a clade of those distributed in North & South parasitic and herb-parasitic groups of genera, were recog- America (Fig. 3). Then, the latter clade split off N. hiratae now nized in the taxonomic system of Braun (1987). For instance, distributed in East Asia. The following scenario may be Cystotheca, Podosphaera, Microsphaera, Uncinula, Typhulochaeta, postulated. Neoe¨rysiphe spp. except for N. galeopsidis split Pleochaeta, and Phyllactinia are tree-parasitic, whilst Blumeria, across the NALB into a European and a North American group. Erysiphe s. l. (containing Golovinomyces and Neoe¨rysiphe), Lev- Then, a part of the North American group migrated into East eillula, and Sphaerotheca are herb-parasitic. The number of host Asia via the BLB. However, an estimation of dating by molec- species was calculated using Table 4 of Amano (1986) based on ular clock suggests that these events occurred after the the assumption that all hosts of tree-parasitic genera and disappearance of the Turgai Strait (Takamatsu et al. 2008a). herb-parasitic genera were actually trees and herbs, respec- Therefore, there might have been additional barriers for plant tively. The ratio of tree hosts was 34.0%, 31.0%, and 9.8% in exchange like large arid areas between Europe and East Asia East Asia, North America, and Europe, respectively. When the after the disappearance of the Turgai Strait. ratio of tree hosts was calculated for each powdery mildew According to Hirata (1969, 1971b) and Amano (1986), the genus, the ratio of hosts of Uncinula was about two times ratio of tree-parasitic to herb-parasitic taxa is high in East Asia higher in East Asia than North America. On the other hand, the ratio of hosts of Microsphaera was higher in North America than in East Asia. In East Asia, 51 Uncinula species infect 28 host genera containing 171 species. In North America, on the other hand, only eight Uncinula species infect seven host genera contain- ing 90 species. Likewise in Europe, only six Uncinula species infect six genera containing 88 species. Therefore, both number of Uncinula species and number of hosts are much more abundant in East Asia than in either North America or Europe. A similar but less conspicuous tendency is also found in Phyllactinia. Since tree-parasitic genera generally have an origin older than that of herb-parasitic genera, the above results suggest that powdery mildews of older origin have flourished in East Asia whilst those of more recent origin have flourished in Europe. In this respect, North America is inter- mediate between the two regions. Powdery mildew genera can also be divided into two other Fig. 2 e Schematic diagram showing comparisons of groups, viz. genera with many species, many hosts and wide powdery mildew flora among East Asia, North America, distributions, and genera with small numbers of species and and Europe. East Asia was isolated from Europe by the hosts, and narrow distributions. The former group consists of Turgai Strait during the Paleogene Period. Europe and East Podosphaera, Microsphaera, Uncinula, Phyllactinia, Erysiphe s. l. Asia were connected to North America by the North (containing Golovinomyces and Neoe¨rysiphe), Leveillula, and Atlantic Land Bridge (NALB) and the Bering Land Bridge Sphaerotheca, and the latter includes Cystotheca, Pleochaeta, (BLB), respectively, in the Paleogene Period. Typhulochaeta, Brasiliomyces, Parauncinula, and Caespitotheca. 82 mycoscience 54 (2013) 75e86

Because Brasiliomyces is a polyphyletic genus it is excluded America. Furthermore, all Phyllactinia species that are sister to from the subsequent consideration. It is noteworthy that all the Leveillula clade are distributed in North & South America, genera belonging to the latter group are tree-parasitic. Cys- and none of them are distributed in Central & West Asia or the totheca, Pleochaeta, and Typhulochaeta are distributed in Asia Mediterranean region (unpublished data). Several explana- and North & South America, but not in Europe (Hirata 1975) tions would be possible to interpret this strange relationship (Fig. 2). Parauncinula and Caespitotheca are distributed in East between Leveillula and its sister group. One explanation could Asia and South America, respectively, but not in Europe. In be as follows: Leveillula originated in North or South America, summary, tree-parasitic genera with a small number of and later it migrated to Central & West Asia or the Mediter- species and hosts, and probably of old origin, are distributed in ranean region where it radiated, whilst the ancestral Leveillula Asia and North & South America, but not in Europe. became extinct in South and North America. Another possible As described above, the major radiation of powdery explanation could be: The closest relatives of Leveillula in mildews occurred in the Paleogene and Neogene periods, Phyllactinia are distributed in Central & West Asia or the a period of climate deterioration and large scale orogeny Mediterranean region. However, we accidentally did not use which formed the Alps, Himalayas, and Rockies. As the earth these species in the current analysis. However, there is a third cooled the broad-leaved deciduous trees growing in high possible explanation that I prefer, viz. an ancestral Phyllactinia latitudes in the Northern Hemisphere moved to middle lati- species originating in the higher latitudes of the Northern tudes and survived in several refugia where speciation Hemisphere migrated southward to North America and occurred. The Western part of North America and Europe also Europe during the global cooling of the world. A part of the may have had a rich Paleogene flora until the beginning of the North American population migrated further to East Asia via Neogene Period when large scale extinction occurred due to the BLB. Leveillula diverged from an ancestral species of Phyl- global cooling and large scale desertification. Finally, most lactinia that migrated to Central & West Asia or the Mediter- Neogene flora became extinct in Europe during the Ice Ages of ranean region due to an adaptation to arid environments. The the Quaternary Period. On the other hand, East Asia became ancestral Phyllactinia species became extinct with its host the largest refugium of this flora in the world due to its warm due to further climate and environmental changes in this and humid climate (Tiffney 1985a,b; Manchester 1999; Wen area. Further investigations are required to evaluate this 1999; Tiffney and Manchester 2001). The Eastern part of hypothesis. North America also functioned as a refugium. Thus both East Asia and North America became refugia for ancestral powdery 3.2.7. Host shift from trees to herbs mildews and would have avoided a possible mass extinction A host shift from trees to herbs may have occurred many of powdery mildews in Europe during the Neogene and times during the evolutionary history of powdery mildews. Quaternary periods. The richer flora of ancestral powdery For instance, it occurred at least twice in the tribe Cys- mildews in East Asia compared with those in North America totheceae (Takamatsu et al. 2000, 2010). One of these may reflects well the difference in the sizes of their refugia. have been a shift of a Podosphaera sp. from the tribe Amyg- daloideae (Rosaceae) to herbs of Scrophulariaceae, leading to 3.2.6. Unexpected phylogeny and distributions of endophytic the creation of sect. Sphaerotheca subsect. Magnicellulatae. genera Another one was a shift from rosaceous trees (tribe Maloideae The genera Phyllactinia, Pleochaeta, Queirozia,andLeveillula, of Rosaceae) to rosaceous herbs, where the pathogen formed which present a monophyletic group within the Erysiphales, the subsect. Sphaerotheca of Podosphaera sect. Sphaerotheca.In are characterized by the formation of endophytic mycelium. the tribe Phyllactinieae, evolution from Phyllactinia (tree The most ancestral genera of this group are Pleochaeta and parasite) to Leveillula (herb parasite) may have occurred only the closely allied Queirozia which form a grade (not clade) at once. It is noteworthy that Leveillula species parasitic to the base of this lineage. Phyllactinia diverged from a part of Asteraceae usually occupy the basal part of the Leveillula tree the Pleochaeta grade, and then Leveillula diverged from (Khodaparast et al. 2001, 2012; Voytyuk et al. 2009). In the tribe Phyllactinia. Erysipheae, herb-parasitic taxa form many small groups Hirata (1968), Amano (1986), and Palti (1988) mentioned the scattered mainly towards the upper half of the Erysipheae tree uneven distribution of Leveillula in the world. Leveillula has (Takamatsu et al. 1999), suggesting that host shift from trees been recorded on many host species in Central & West Asia to herbs occurred many times independently in this tribe. A and the Mediterranean region, but much fewer hosts are comprehensive phylogenetic tree of this lineage is urgently recorded in East Asia, South Asia, North & South America, required to discuss evolution of this group. Oceania, and Northern Europe. Hirata (1968) and Amano (1986) Among the three genera in the tribe Golovinomyceteae, speculated that the geographic origin of Leveillula is Central & Golovinomyces and Neoe¨rysiphe mainly infect herbs and only East Asia or the Mediterranean region, from where Leveillula is Arthrocladiella infects shrubs. Because Arthrocladiella does not in the process of expanding its distribution in the world. Our occupy the basal part of this lineage, an ancestral tree- molecular phylogenetic analysis indicated that Leveillula has parasitic taxon of the Golovinomyceteae is not known yet. the most recent origin in Erysiphaceae, thus supporting this As mentioned above, Asteraceae may be closely related to the speculation (Takamatsu et al. 2008b). early evolution of Golovinomyces (Matsuda and Takamatsu The closest relative of Leveillula is Phyllactinia adesmiae 2003). Neoe¨rysiphe is a relatively small genus infecting 336 (Khodaparast et al. 2012) recorded only in Argentina (South host species in eight families (Amano 1986). Of these host America). The next closest relative is Ovulariopsis obclavata species, 267 (79.5%) belong to the Lamiaceae, suggesting Neo- (anamorph of Phyllactinia) also recorded only in South e¨rysiphe has a close relationship with this family. In its mycoscience 54 (2013) 75e86 83

phylogenetic tree, a group of isolates from Lamiaceae and USA. He was aware that G. orontii was recorded on many (N. galeopsidis group) split firstly from a group consisting of host species in the USA whilst Po. xanthii was recorded in isolates from other plant families (Takamatsu et al. 2008a; Japan but not in the USA. In order to determine whether there Heluta et al. 2010). However, genetic diversity of the was a similar tendency between East Asia and Europe, we N. galeopsidis group is quite low, that is, most isolates have an compared the number of hosts of G. orontii and Po. xanthii identical ITS sequence and maximum substitutions within among East Asia, North America, and Europe based on Ama- the group are only 4 bases. This low genetic divergence of no’s book (1986). About four times more host species of Golo- N. galeopsidis suggests that infection of Neoe¨rysiphe to Lam- vinomyces are recorded in Europe than in Asia either in total or iaceae was a relatively recent event despite Neoe¨rysiphe having as hosts of Asteraceae (Table 1). Thus Golovinomyces seems to such a high proportion of Lamiaceae hosts. Furthermore, be more closely associated with Europe than with East Asia. many Asteraceous hosts are included in the primary group On the other hand, the difference is not so large in the number which forms several distinct clades with a high genetic of host species of P. sect. Sphaerotheca subsect. Magnicellulatae diversity depending on the regions where they were collected. (mainly Po. xanthii) between these two regions, although From these results, Asteraceae seems to be closely related to Europe still has the most species. the early evolution of Neoe¨rysiphe as well as to the early Since Blumeria may have originated due to a host shift from evolution of Leveillula and Golovinomyces. Voytyuk et al. (2004, a dicotyledon to an ancestor of Pooideae, the geographic origin 2006) and Heluta et al. (2010) pointed out that many Neo- of the latter might be relevant to the geographic origin of e¨rysiphe specimens on Asteraceae may have been erroneously Blumeria. Several reports discussing the origin of Pooideae deposited in European herbaria as Golovinomyces specimens. mostly agree that the subfamily originated in the region from Therefore, the number of Asteraceous hosts of Neoe¨rysiphe the Mediterranean to Southwest Asia (Blattner 2006; Inda et al. should increase following re-assessments of herbarium 2008) and so these regions could be the geographic origin of specimens. It is noteworthy that such erroneous identification Blumeria. occurred mainly in the Mediterranean region. As mentioned above, the early evolution of the three herb- Blumeria is the only genus in the Erysiphaceae that infects parasitic genera, Leveillula, Golovinomyces, and Neoe¨rysiphe monocotyledons. Host range of this genus is restricted to the seems, like Blumeria, to be closely associated with the region subfamily Pooideae of Poaceae. Apart from Leveillula allii, all from the Mediterranean to Central & West Asia. One notable other powdery mildew genera infect dicotyledons. So, it is fact is that none of these share an ancestor with tree-parasitic likely that the ancestor of Blumeria underwent a host shift taxa. In the tribes Erysipheae, Cystotheceae and Phyllacti- from dicotyledons to Pooideae. The Blumeria lineage consists nieae, both tree-parasitic taxa and herb-parasitic taxa are only of a single species B. graminis and is only distantly related represented in their respective lineages, although tree- to all other powdery mildew lineages. Thus, the ancestral taxa parasitic taxa usually occupy the basal parts. However, with leading to Blumeria are presently unknown. the exception of Arthrocladiella in the tribe Golovinomyceteae there is no other tree-parasitic taxon in this tribe or in Blu- 3.2.8. Geographic origin of herb-parasitic powdery mildews merieae. In the case of Leveillula, it is unclear whether it shares As discussed above, the family Asteraceae may be closely a direct ancestor with its sister, Ph. adesmiae. Extinction of related to the early evolution of at least three herb-parasitic woody plants in this region could be a possible explanation for genera, viz. Golovinomyces, Leveillula, and Neoe¨rysiphe. These the lack of tree hosts in these lineages. Central & West Asia were unexpected results because the Asteraceae is of rela- was a region surrounded by the Turgai Strait and Tethys Sea in tively recent origin among Angiosperms (Bremer 1994). South the Paleogene Period and vegetated by Pterocarya, Betula, Alnus, America is considered to be the geographic origin of the Asteraceae (Bremer 1994). Then, this family expanded into the Northern Hemisphere and underwent explosive radiation and e dispersal during or near the Oligocene and Miocene of the Table 1 Number of host species of the genus Golovinomyces and of Podosphaera sect. Sphaerotheca Paleogene and Neogene periods (Kim et al. 1998). The tribe subsect. Magnicellulatae in East Asia, North America, and Cardueae was the first major group of the Asteraceae to Europe. disperse throughout the Northern Hemisphere (Bremer 1994; Number of host speciesa Kim and Jansen 1995), where it differentiated into genera with hundreds of species in the Mediterranean region and Central World East Asia North Europe & West Asia (Bremer 1994). America Hirata (1968) noted that Podosphaera xanthii (currently in Golovinomyces subsect. Magnicellulatae of Podosphaera sect. Sphaerotheca) Total host 2283 287 (12.6%) 685 (30.0%) 1108 (48.5%) [ ¼ S. fuliginea in Hirata (1968)] is commonly reported as species a causal agent of powdery mildew of cucumber and cucurbit Asteraceae 1242 167 (13.4%) 401 (32.3%) 642 (51.7%) in Japan, whereas Golovinomyces orontii [ ¼ Erysiphe cichor- host species P. sect. Sph. subsect Magnicellulatae acearum in Hirata (1968)] is recorded as the main cause of Total host 1110 291 (26.2%) 228 (20.5%) 426 (38.4%) powdery mildew of Cucurbitaceae in many other countries. species The latter was first found in Japan in 2002, and within several Asteraceae 499 152 (30.5%) 110 (22.0%) 191 (38.3%) years expanded to cover all parts of the country (Hoshi et al. host species 2009; Uchida et al. 2009). Hirata (1955, 1968) compared a Numbers based on the list of Amano (1986). records of powdery mildews and their hosts between Japan 84 mycoscience 54 (2013) 75e86

Salix, Nyssa, and so forth (Tiffney and Manchester 2001). The phylogenetic relationships of powdery mildews, multigene subsequent Alpine and Himalayan orogenies shut off the analyses including protein-coding regions are necessary to southern source of moisture, allowing this area to become arid evaluate the phylogeny of powdery mildews more accu- during the Miocene of the Neogene Period when the diversity rately. Multigene analyses of powdery mildews have of Angiosperms decreased. However, information on the currently been initiated by several research groups. I hope changes in Central & West Asian flora is not adequate to that the discussion conducted in this review will be discuss this subject in detail. Further investigations are a trigger for further discussion on the origin and evolution required. of the powdery mildews. With regard to P. sect. Sph. subsect. Magnicellulatae, Hirata (1969) found a relatively high proportion of this subsect. as well as B. graminis amongst powdery mildews in latitudes above 60N. Junell (1967) also reported frequent distribution of Disclosure this subsect. in the northern part of Sweden. Worldwide it infects 206 species of Scrophulariaceae, of which 113 (54.9%) All the experiments undertaken in this study comply with the were recorded in Europe. On the other hand, only 27 (13.1%) current laws of Japan. host species were recorded in East Asia and only 41 (19.9%) in North America. Therefore, the area from Europe to Central Asia might be closely associated with the early evolution of subsect. Magnicellulatae. Other herb-parasitic powdery mildews belonging to Ery- Acknowledgments siphe sect. Erysiphe in the tribe Erysipheae show a complicated The author wishes to thank Drs Roger Cook and Uwe Braun for polyphyly, which suggests that host shifts to herbaceous critical reading of the manuscript and helpful comments. The plants have occurred many times at different places in this author also thanks the late Prof. Koji Amano (formerly lineage. There are not enough phylogenetic analyses of the K. Hirata) for providing an opportunity to work on this inter- tribe Erysipheae to discuss the origin of herb-parasitic taxa in esting research subject. This study was supported, in part, by this group. a Grant-in-Aid for Scientific Research No. 23580061 from the Japan Society for the Promotion of Science. 4. Prospects

Discussions in this review are focused on two significant references subjects: 1) estimation of evolutionary dating using molecular clock newly designed for powdery mildews, and 2) geographic origin of powdery mildews. Calibration of evolutionary dating Ainsworth GC, James PW, Hawksworth DL, 1971. Ainsworth and is essential to discuss evolution of organisms like powdery Bisby’s dictionary of the fungi including the lichens, 6th edn. CAB mildews whose life cycle is completely dependent on other International, Kew. Alexopoulos CJ, Mims CW, 1979. Introductory mycology, 3rd edn. organisms. Fortunately, we were able to find a calibration John Wiley & Sons, New York. e point for a molecular clock based on host parasite relation- Alexopoulos CJ, Mims CW, Blackwell M, 1996. Introductory ships, which we used in a newly designed molecular clock mycology, 4th edn. John Wiley & sons, New York. employing ITS and 28S rDNA D1/D2 regions. This clock Amano K, 1986. Host range and geographical distribution of the enabled us to discuss various aspects of the evolution of powdery mildew fungi. Japan Scientific Societies Press, Tokyo. powdery mildews. It is necessary to re-evaluate these aspects Amano K, 1992. Notes on the host range and geographical distribution of Podosphaera. Transactions of the Mycological from a different viewpoint or using a new molecular clock Society of Japan 33: 139e148. constructed with a different base. Amano K, 2002. Notes on host range and geographical The discussion on the geographic origin of powdery distribution of Uncinula (in Japanese). Nippon Kingakukai Kaiho mildews was challenging due to the scarcity of usable 43: 127e130. evidence. Thus, the present discussion has to be mainly based Arnaud G, 1921. E´tude sur les champignons parasites. Annales des on a lot of speculation, the evaluation of which requires E´piphyties 7: 1e115. further molecular data. The following approaches are neces- Barrett PM, Willis KJ, 2001. Did dinosaurs invent flowers? Dinosaureangiosperm coevolution revisited. Biological Review sary in future to improve the situation: 76: 411e447. Batra L, Segal R, Baxter R, 1964. A new Middle Pennsylvanian 1) Comprehensive molecular phylogenetic analyses of more fossil fungus. American Journal of Botany 51: 991e995. (maybe all) powdery mildews. Because the nucleotide Berbee ML, Taylor JW, 1993. Dating the evolutionary radiation of sequence data so far determined are only from a part of the the true fungi. Canadian Journal of Botany 71: 1114e1127. powdery mildew genome, many more sequence data are Berbee ML, Taylor JW, 2001. Fungal molecular evolution: gene required to infer comprehensive phylogenetic relationships trees and geologic time. In: McLaughlin DJ, Mclaughlin EG, Lemke PA (eds), The Mycota VII. Systematics and evolution, part B. of this fungal group on a worldwide basis. Springer-Verlag, Berlin, pp 229e245. 2) Multigene phylogenetic analyses. Although ribosomal DNA Blattner FR, 2006. Multiple intercontinental dispersals shaped the regions like ITS or 28S rDNA D1/D2 regions have been distribution area of Hordeum (Poaceae). New Phytologist 169: extremely useful for the broad scale elucidation of 603e614. mycoscience 54 (2013) 75e86 85

Blumer S, 1933. Die Erysiphaceen Mitteleuropas unter besonderer Hirata K, 1980. Host plants of powdery mildew fungi collected at Beru¨ cksichtigung der Schweiz. Beitra¨ge zur Kryptogamenflora the Royal Botanic Gardens, Kew, England, during August and der Schweiz 7: 1e483. September in 1978. Transactions of the Mycological Society of Braun U, 1987. A monograph of the Erysiphales (powdery Japan 21: 245e258. mildews). Beihefte zur Nova Hedwigia 89: 1e700. Hoshi H, Sato Y, Horie H, 2009. Current status of cucumber powdery Braun U, 1995. The powdery mildews (Erysiphales) of Europe.G. mildew caused by Oidium subgenus Reticuloidium in Tokyo, Fischer Verlag, Jena. Japan (in Japanese). Japanese Journal of Phytopathology 75: 21e28. Braun U, Cook RTA, Inman AJ, Shin HD, 2002. The taxonomy of Inda LA, Segarra-Moragues JG, Muller J, Peterson PM, Catalan P, the powdery mildew fungi. In: Belanger R, Dik AJ, Bushnell WR 2008. Dated historical biogeography of the temperate Loliinae (eds), The powdery mildews: a comprehensive treatise. APS Press, (Poaceae, Pooideae) grasses in the northern and southern St. Paul, pp 13e54. hemispheres. Molecular Phylogenetics and Evolution 46: 932e957. Braun U, Cook RTA, 2012. Taxonomic manual of Erysiphales (powdery Junell L, 1967. Erysiphaceae of Sweden. Symbolae Botanicae mildews). In: CBS Biodiversity Series No 11. CBS, Utrecht. Upsalienses 14: 1e117. Bremer K, 1994. Asteraceae cladistics & classification. Timber Press, Katumoto K, 1973. Notes on the genera Lanomyces Ga¨um. and Portland, Oregon. Cystotheca Berk. et Curt. Reports of the Tottori Mycological Institute Bremer K, Gustafsson MHG, 1997. East Gondwana ancestry of the 10: 437e446. sunflower alliance of families. In: Proceedings of the National Khodaparast SA, Takamatsu S, Hedjaroude GA, 2001. Academy of Sciences of the United States of America 94, Phylogenetic structure of the genus Leveillula (Erysiphales: pp 9188e9190. Erysiphaceae) inferred from the nucleotide sequences of the Gandolfo MA, Nixon KC, Crepet WL, Stevenson DW, Friis EM, rDNA ITS region with special references to the Leveillula taurica 1998. Oldest known fossils of monocotyledons. Nature 394: species complex. Mycological Research 105: 909e918. 532e533. Khodaparast SA, Takamatsu S, Harada M, Abbasi M, Samadi S, Glawe DA, 2008. The powdery mildews: a review of the world’s 2012. Additional rDNA ITS sequences and its phylogenetic most familiar (yet poorly known) plant pathogens. Annual consequences for the genus Leveillula with emphasis on Review of Phytopathology 46: 27e51. conidium morphology. Mycological Progress 11: 741e752. Havrylenko M, 1993. Descriptions of new taxa of Erysiphaceae Kim HG, Keeley SC, Vroom PS, Jansen RK, 1998. Molecular from Argentina. Mycotaxon 49: 257e267. evidence for an African origin of the Hawaiian endemic Havrylenko M, Takamatsu S, 2003. Erysiphe patagoniaca: a new Hesperomannia (Asteraceae). Proceedings of the National Academy species of Erysiphe sect. Uncinula from Patagonia, Argentina. of Sciences of the United States of America 95: 15440e15445. Mycoscience 44: 149e151. Kim KJ, Jansen RK, 1995. ndhF sequence evolution and the major Heluta VP, 1992. Gipoteza pro pokhodzhennya ta migrazii gribiv clades in the sunflower family. Proceedings of the National poryadku Erysiphales (in Ukranian). Ukrayins’kyi Botanichnyi Academy of Sciences of the United States of America 92: Zhurnal 49: 5e14. 10379e10383. Heluta V, Takamatsu S, Harada M, Voytyuk S, 2010. Molecular Kimura M, 1980. A simple method for estimating evolutionary phylogeny and taxonomy of Eurasian Neoe¨rysiphe species rates of base substitutions through comparative studies of infecting Asteraceae and Geranium. Persoonia 24: 81e92. nucleotide sequences. Journal of Molecular Evolution 16: 111e120. Hermansen JE, Torp U, Prahm LP, 1978. Studies of transport of cereal Manchester SR, 1999. Biogeographical relationships of North mildew and rust fungus across the North Sea. Grana 17:1e46. American Tertiary floras. Annals of the Missouri Botanical Garden Hirata K, 1955. Comparison of powdery mildews and their host 86: 472e522. plants of Japan and the United States of America (in Japanese). Matsuda S, Takamatsu S, 2003. Evolution of hosteparasite Bulletin of the Faculty of Agriculture, Niigata University 7: 146e152. relationships of Golovinomyces (Ascomycete: Erysiphaceae) Hirata K, 1966. Host range and geographical distribution of the inferred from nuclear rDNA sequences. Molecular Phylogenetics powdery mildews. Faculty of Agriculture, Niigata University, and Evolution 27: 314e327. Niigata. Meeboon J, Takamatsu S, 2012. Erysiphe havrylenkoana and E. Hirata K, 1968. Notes on host range and geographic distribution of prunastri var. japonica: a new species and a new variety of the powdery mildew fungi. Transactions of the Mycological Erysiphe sect. Uncinula (Erysiphaceae, Ascomycota). Mycological Society of Japan 9: 73e88. Progress; http://dx.doi.org/10.1007/s11557-012-0832-z. Hirata K, 1969. Notes on host range and geographic distribution of Moran NA, Munson MA, Baumann P, Ishikawa H, 1993. A the powdery mildew fungi II. Transactions of the Mycological molecular clock in endosymbiotic bacteria is calibrated Society of Japan 10: 47e72. using the insect hosts. Proceedings of the Royal Society B 253: Hirata K, 1971a. Host range and geographic distribution of 167e171. Erysiphaceae as viewed from the families of Angiosperms, Mori Y, Sato Y, Takamatsu S, 2000a. Evolutionary analysis of the and comparison with Meliolineae. Sydowia 25: 100e118. powdery mildew fungi using nucleotide sequences of the Hirata K, 1971b. Notes on host range and geographic distribution nuclear ribosomal DNA. Mycologia 92: 74e93. of the powdery mildew fungi III. Transactions of the Mycological Mori Y, Sato Y, Takamatsu S, 2000b. Molecular phylogeny and Society of Japan 12: 1e13. radiation time of Erysiphales inferred from the nuclear Hirata K, 1972. Notes on host range and geographic distribution of ribosomal DNA sequences. Mycoscience 41: 437e447. the powdery mildew fungi IV. Transactions of the Mycological Neger FW, 1901. Beitra¨ge zur Biologie der Erysipheen. Flora 88: Society of Japan 13: 1e21. 333e370. Hirata K, 1975. Notes on host range and geographic distribution of Niinomi S, Takamatsu S, Havrylenko M, 2008. Molecular data do the powdery mildew fungi V. Uneven distribution of the not support a Southern Hemisphere base of Nothofagus powdery mildew fungi in the temperate zone of the northern powdery mildews. Mycologia 100: 710e720. hemisphere, especially of Eurasia. Transactions of the Oberhaensli S, Parlange F, Buchmann JP, Jenny FH, Abbott JC, Mycological Society of Japan 16: 113e127. Burgis TA, Spanu PD, Keller B, Wicker T, 2011. Comparative Hirata K, 1976. Notes on host range and geographic distribution of sequence analysis of wheat and barley powdery mildew fungi the powdery mildew fungi VI. Distribution of the hosts of reveals gene colinearity, dates divergence and indicates powdery mildew fungi in the families of angiosperms. hostepathogen co-evolution. Fungal Genetics and Biology 48: Transactions of the Mycological Society of Japan 17: 35e62. 327e334. 86 mycoscience 54 (2013) 75e86

Palti J, 1988. The Leveillula mildew. Botanical Review 54: 423e535. molecular phylogenetic analysis and evolution of the genus Peterson KR, Pfister DH, 2010. Phylogeny of Cyttaria inferred from Phyllactinia (Ascomycota: Erysiphales) and its allied genera. nuclear and mitochondrial sequence and morphological data. Mycological Research 112: 299e315. Mycologia 102: 1398e1416. Takamatsu S, Niinomi S, Harada M, Havrylenko M, 2010. Peterson KR, Pfister DH, Bell CD, 2010. Cophylogeny and Molecular phylogenetic analyses reveal a close evolutionary biogeography of the fungal parasite Cyttaria and its host relationship between Podosphaera (Erysiphales: Erysiphaceae) Nothofagus, southern beech. Mycologia 102: 1417e1425. and its rosaceous hosts. Persoonia 24: 38e48. Pirozynski KA, 1976. Fossil fungi. Annual Review of Phytopathology Taylor TN, Krings M, 2005. Fossil microorganisms and land 14: 237e246. plants: associations and interactions. Symbiosis 40: 119e135. Pryce-Jones E, Carver TLW, Gurr SJ, 1999. The roles of cellulase Taylor TN, Krings M, Klavins SD, Taylor EL, 2005. Protoascon enzymes and mechanical force in host penetration by Erysiphe missouriensis, a complex fossil microfungus revisited. graminis f. sp hordei. Physiological and Molecular Plant Pathology Mycologia 97: 725e729. 55: 175e182. Tiffney BH, 1985a. Perspectives on the origin of the floristic Raymond J, 1927. Le blanc du cheˆne.Annales des Epiphyties 13: 94e129. similarity between eastern Asia and eastern North America. Saenz GS, Taylor JW, Gargas A, 1994. 18S rRNA gene sequences Journal of the Arnold Arboretum 66: 73e94. and supraordinal classification of the Erysiphales. Mycologia Tiffney BH, 1985b. The Eocene North Atlantic Land Bridge: its 86: 212e216. importance in tertiary and modern phytogeography of the Schnathorst WC, 1959. Spread and life cycle of the lettuce northern hemisphere. Journal of the Arnold Arboretum 66: 243e273. powdery mildew fungus. Phytopathology 49: 464e468. Tiffney BH, 2000. Geographic and climatic influence on the Simon L, Bousquet J, Le´vesque RC, Lalonde M, 1993. Origin and Cretaceous and Tertiary history of Euramerican floristic diversification of endomycorrhizal fungi and coincidence with similarity. Acta Universitatis Carolinae, Geologica 44: 5e16. vascular land plants. Nature 363: 67e69. Tiffney BH, Barghoorn ES, 1974. The fossil record of fungi. Occasional Stubblefield SP, Taylor TN, 1983. Studies of Paleozoic Fungi. I. The Papers of the Farlow Herbarium of Cryptogamic Botany 7: 1e42. Structure and Organization of Traquairia (Ascomycota). Tiffney BH, Manchester SR, 2001. The use of geological and American Journal of Botany 70: 387e399. paleontological evidence in evaluating plant phylogeographic Su ZH, Tominaga O, Okamoto M, Osawa S, 1998. Origin and hypotheses in the Northern Hemisphere Tertiary. International diversification of hindwingless Damaster ground beetles Journal of Plant Science 162 (suppl): S3eS17. within the Japanese islands as deduced from mitochondrial Tsuneda A, Currah RS, 2004. Ascomatal morphogenesis in ND5 gene sequences (Coleoptera, Carabidae). Molecular Biology Myxotrichum arcticum supports the derivation of the and Evolution 15: 1026e1039. Myxotrichaceae from a discomycetous ancestor. Mycologia 96: Sugiyama M, Ohara A, Mikawa T, 1999. Molecular phylogeny of 627e635. onygenalean fungi based on small subunit ribosomal DNA Uchida K, Takamatsu S, Matsuda S, So K, Sato Y, 2009. (SSU rDNA) sequences. Mycoscience 40: 251e258. Morphological and molecular characterization of Oidium Takahashi M, 2005. Vegetational consequences of Cretaceous and subgenus Reticuloidium (powdery mildew) newly occurred on Paleogene flora: Early radiation of Angiosperms in Cretaceous cucumber in Japan. Journal of General Plant Pathology 75: age and biogeographical distribution of Early Tertiary Flora (in 92e100. Japanese). Journal of the Japanese Association for Petroleum Voytyuk S, [as Voityuk], Heluta V, Nevo E, 2004. Neoe¨rysiphe Technology 70: 37e46. cumminsiana (Erysiphales, Eumycota), a new powdery mildew Takahashi M, 2006. Origin and early evolution of the Angiosperms (in fungus in Israel. Flora Mediterranea 14: 267e273. Japanese). Hokkaido University Press, Sapporo. Voytyuk SO, Heluta VP, Wasser SP, Nevo E, 2006. Genus Takamatsu S, Hirata T, Sato Y, Nomura Y, 1999. Phylogenetic Neoe¨rysiphe in Israel: species composition, host range and relationships of Microsphaera and Erysiphe section Erysiphe distribution. Mycotaxon 97: 247e256. (powdery mildews) inferred from the rDNA ITS sequences. Voytyuk SO, Heluta VP, Wasser SP, Nevo E, Takamatsu S, 2009. Mycoscience 40: 259e268. Biodiversity of the powdery mildew fungi (Erysiphales, Ascomycota) Takamatsu S, Hirata T, Sato Y, 2000. A parasitic transition from of Israel. Gantner Verlag, Ruggell. trees to herbs occurred at least twice in tribe Cystotheceae Wang Z, Binder M, Schoch CL, Johnston PR, Spatafora JW, (Erysiphaceae): evidence from nuclear ribosomal DNA. Hibbett DS, 2006a. Evolution of helotialean fungi Mycological Research 104: 1304e1311. (Leotiomycetes, Pezizomycotina): a nuclear rDNA phylogeny. Takamatsu S, 2004. Phylogeny and evolution of the powdery Molecular Phylogenetics and Evolution 41: 295e312. mildew fungi (Erysiphales, Ascomycota) inferred from nuclear Wang Z, Johnston PR, Takamatsu S, Spatafora JW, Hibbett DS, ribosomal DNA sequences. Mycoscience 45: 147e157. 2006b. Toward a phylogenetic classification of the Takamatsu S, Matsuda S, 2004. Estimation of molecular clocks for Leotiomycetes based on rDNA data. Mycologia 98: 1065e1075. ITS and 28S rDNA in Erysiphales. Mycoscience 45: 340e344. Webster J, 1980. Introduction of fungi, 2nd edn. Cambridge Takamatsu S, Niinomi S, Cabrera de A´ lvarez MG, A´ lvarez RE, University Press, Cambridge. HavrylenkoM, Braun U,2005. Caespitothecagen. nov., anancestral Wen J, 1999. Evolution of Eastern Asian and Eastern North genus in the Erysiphales. Mycological Research 109: 903e911. American disjunct distributions in flowering plants. Annual Takamatsu S, Matsuda S, Niinomi S, Havrylenko M, 2006. Review of Ecology and Systematics 30: 421e455. Molecular phylogeny supports a northern hemisphere origin Wikstro¨ m N, Savolainen V, Chase MW, 2001. Evolution of the of Golovinomyces (Ascomycota: Erysiphales). Mycological angiosperms: calibrating the family tree. Proceedings of the Research 110: 1093e1101. Royal Society B 268: 2211e2220. Takamatsu S, Havrylenko M, Wolcan SM, Matsuda S, Niinomi S, Yarwood CE, 1973. Pyrenomycetes: Erysiphales. In: Ainsworth GC, 2008a. Molecular phylogeny and evolution of the genus Sparrow FK, Sussman AS (eds). The fungi: an advanced treatise, Neoe¨rysiphe (Erysiphaceae, Ascomycota). Mycological Research vol. IV. A. Academic Press, New York, pp 71e86. 112: 639e649. Yarwood CE, 1978. History and taxonomy of the powdery Takamatsu S, Inagaki M, Niinomi S, Khodaparast SA, Shin HD, mildews. In: Spencer DM (ed) The powdery mildews. Academic Grigaliunaite B, Havrylenko M, 2008b. Comprehensive Press, New York, pp 1e37.